Loading...

Design and Development of a Portable Electrocardiograph Capable of Arrhythmia Detection Based on Deep Learning

Kamandi, Kian | 2024

0 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 57813 (05)
  4. University: Sharif University of Technology
  5. Department: Electrical Engineering
  6. Advisor(s): Farhadi, Alireza
  7. Abstract:
  8. According to the World Health Organization, approximately 17.9 million people worldwide lose their lives due to cardiovascular diseases. Continuous monitoring of electrocardiogram (ECG) signals enables early detection of cardiovascular diseases and plays a key role in reducing the risks associated with these conditions. In this thesis, a single-lead, low-cost, portable, and accessible system suitable for intelligent ECG signal monitoring at home with simple and inexpensive hardware has been developed. This system consists of two parts: hardware and software (an application) that can be installed on personal computers running Windows. In the application section of this system, an artificial intelligence model named ECG-EffNetX, based on convolutional neural networks, has been developed. This network has been trained using the MIT-BIH database and achieved 99.42% cross-validation accuracy and 99.40% test accuracy. Users can view their ECG in real-time through a Windows-based graphical interface on their computer and perform precise analyses of their heart condition with the help of this model. Since explainability of the model is vital in health applications, the developed model has been made fully explainable using the Gradient-weighted Class Activation Mapping algorithm. This capability ensures that the developed system is highly reliable. This system has been applied to multiple individuals, and the results of the system's reports have been compared with their cardiovascular disease histories, ensuring the system's desirable performance
  9. Keywords:
  10. Electrocardiogram ; Deep Learning ; Explainable Artificial Intelligence ; Convolutional Neural Network ; Smart Monitoring ; Digital Health ; Gradient-Weighted Class Activation Mapping (Grad-CAM) ; Early Detection ; Cardiovascular Patients

 Digital Object List

 Bookmark

No TOC